node deployment strategies and coverage prediction in 3d wireless sensor network … ·...

Advances in Computational Sciences and Technology

ISSN 0973-6107 Volume 10, Number 8 (2017) pp. 2243-2255

© Research India Publications

http://www.ripublication.com

Node Deployment Strategies and Coverage Prediction

in 3D Wireless Sensor Network with Scheduling

Anvesha Katti

School of Computer and System Sciences Jawaharlal Nehru University, New Delhi, India

D.K.Lobiyal

School of Computer and System Sciences Jawaharlal Nehru University, New Delhi, India

Abstract

Wireless sensor network (WSN) is a wireless network of spatially distributed

autonomous devices using sensors to monitor physical conditions.

Deployment of sensor nodes is a critical issue in WSN as it affects coverage

and connectivity of the network. Coverage in wireless sensor networks is a

measure of how well and for how long the sensors are able to observe the

physical space. In this paper, we propose different sensor node deployment

strategies for 3D WSN for maximum coverage prediction. We do a

comparative study of 3D sensor node deployment strategies for coverage

prediction. We also propose a scheduling algorithm to minimize the number of

sensor nodes used in coverage prediction. Our study also gives a comparison

between various proposed 3D sensor node deployment schemes along with the

number of sensor nodes to be used in each case.

Keywords: wireless sensor networks, sensor node deployment, coverage,

connectivity

I. INTRODUCTION

Wireless sensor networks (WSN), consist of spatially distributed autonomous sensors

which monitor physical phenomenon and pass on the data collectively to a data

2244 Anvesha Katti and D.K.Lobiyal

collector called a sink. The WSN consist of a few to several hundred sensor nodes,

which are capable of sensing in any environmental conditions. Cost constraints and

detection possibility of sensor nodes have been one of the most researched areas of

WSN 12. WSNs find many applications in environmental observation and forecasting

systems, habitat monitoring, intrusion detection and tracking, seismic monitoring, etc.3

Every sensor in a WSN has a limited sensing range and the union of the sensing

ranges of all sensors reflects how well the area of sensor field is monitored known as

the coverage area 4.Deployment of sensors is a major aspect influencing coverage. The

deployment of a WSN affects many of its metrics such as coverage, connectivity and

lifetime.

There are mainly two types of sensor node deployment: deterministic deployment and

random deployment. Deterministic deployment is used where uniform sensing is

needed. Random deployment is normally used in case of inaccessible terrains, disaster

areas and war zones.

In random deployment, sensors are usually scattered for example air dropped 5.

Deterministic deployment is selectively deciding the locations of the sensors for

uniform coverage by optimizing one or more parameters. Deterministic deployment

finds applications in border surveillance, intrusion detection, and structural healthcare

among others 6.

In this paper, we present a novel deterministic sensor node deployment scheme of 3D

WSN consisting of prism deployment, pyramid deployment, cube deployment and

hexagonal prism deployment along with finding the coverage prediction. We also

propose a scheduling algorithm which will help increase the lifetime of the sensor

nodes by switching off nodes and saving energy.

This paper is organized as follows: in section 2 we present the research already

reported in literature, in section 3 we address various types of deployments of sensor

nodes in 3D WSN and also find the coverage prediction for each of them along with

the number of sensor nodes required for each type of deployment. The results are

explained in section 4 and we conclude in section 5.

II. LITERATURE REVIEW

Good coverage and connectivity in WSN depends a lot on sensor node deployment.

This has been an active area of research with the aim of optimizing parameters

revolving around lifetime, cost, coverage and connectivity.

In 7, the authors present deployment approaches of WSNs optimizing parameters such

as energy consumption and obstacle adaptability using artificial potential field and

computational geometry techniques. Liu and He 8 propose an Ant Colony optimization

Node Deployment Strategies and Coverage Prediction in 3D Wireless Sensor… 2245

with greedy deployment solution for maximizing coverage in grid based WSN.

Authors maximize coverage while minimizing sensor movement in 9 using a complex

algorithm. In the deployment algorithm introduced in 10, each node will communicate

with its neighbors and tell them to move away until they are at a distance which

maximizes coverage while maintaining connectivity.

III. COVERAGE PREDICTION FOR NODE DEPLOYMENTS 3D WSN

ENVIRONMENT

Coverage defines monitoring environmental conditions or changes by sensors

deployed in a certain region. Every sensor in a WSN has a limited sensing range and

the union of the sensing ranges of all sensors is known as the network sensing

coverage reflecting how well the area of sensor field is monitored. One of the major

factors affecting coverage is an effective deployment strategy. Coverage measures how

well each point in the sensing field is covered by sensors. A sensor network

deployment can usually be classified as either a regular deployment or random

deployment as described earlier. We develop a deployment strategy for regular

placement of sensor nodes and assume the sensor nodes are static, i.e. they stay in the

same place once they are deployed. As compared to random deployment, regular

deployment of sensors in WSNs provide better sensing coverage and a higher degree

of connectivity. Good deployment helps us to place the nodes in a manner to maximize

coverage. Coverage prediction for a sensor node is the ratio of the volume covered by

the node to the whole volume of deployment. 11

We derive the coverage prediction for different deployment strategies such as a prism,

pyramid, cube, and hexagonal prism type of deployment and also find the number of

sensors used in each case for coverage prediction. We also give a comparison of the

different deployment strategies with the number of sensor nodes. We start by

describing the disk sensing model which is one of the most widely used models.

A. Coverage using Disk Sensing Model

The sensor has a constant sensing range r with volume 𝑎 = 𝜋𝑟3. If a target node lies in

the sensing range of a sensor it is said to be covered. Probability of target detection is

defined as the ratio of sensing volume to network volume expressed as Pd=v/V where

V is network volume and 𝑁 is the number of sensor nodes deployed uniformly. The

probability of target detection Pc by at least one of the 𝑁 sensors can be expressed as

(1)


By applying the equality approximation as n is very large, the stated

equation can be rewritten as

(2)

This model is generally used in comparison with different probabilistic sensing

models12.

B. Prism Sensor Node Deployment Strategy

Fig.1 Prism type sensor node deployment

We take a cube sensing field of side a with the cube being divided into small

equilateral prism regions. Assumptions include sensors at the end points of the prism.

Sensing radius is the radius of the sphere at the end points of the sensor node (r) and

varies between 0 and Rmax with Rmax being the maximum radius and 0 being the

minimum. It is assumed that the sensors placed at the endpoints have equal sensing

range depicted by the radius of the sphere. Coverage Prediction is the ratio of the

sensing volume to total volume. For prism sensor deployment this is expressed as

(3)

C. Cube Node Deployment

Sensor

r, Rm(sensing radius)

Fig.2. Cube type sensor node deployment


In fig.2 we take a cube sensing field of side a with the cube being divided into small

equilateral cube regions. Assumptions include sensors at the end points of the cube.

Sensing radius is the radius of the sphere at the end points of the sensor node (r) and

varies between 0 and Rmax with Rmax being the maximum radius and 0 being the



sensing volume to total volume. For cube sensor deployment this is expressed as

(4)

C. Hexagonal Prism Node Deployment

sensorr, Rmax(sensing

radius)

Fig.3. Hexagonal Prism Node Deployment


equilateral hexagonal prism regions. Assumptions include sensors at the end points of

the hexagonal prism. Sensing radius is the radius of the sphere at the end points of the

sensor node (r) and varies between 0 and Rmax with Rmax being the maximum radius

and 0 being the minimum. It is assumed that the sensors placed at the endpoints have

equal sensing range depicted by the radius of the sphere. Coverage Prediction is

defined as the ratio of the sensing volume to total volume. For hexagonal prism sensor

deployment, this is expressed as

(5)

E. Pyramid Node Deployment

sensorr, Rmax (sensing radius)

Fig.4. Pyramid Node Deployment



equilateral pyramid regions. Assumptions include sensors at the end points of the

pyramid. Sensing radius is the radius of the sphere at the end points of the sensor node

(r) and varies between 0 and Rmax with Rmax being the maximum radius and 0 being the



sensing volume to total volume. For pyramid sensor deployment this is expressed as

(6)

F. Sensor Node Estimation

By putting r=Rmax in (3), (4), (5) and (6) the maximum coverage prediction for prism

is 2π/3√3, the maximum coverage prediction for cube is π/3, the maximum coverage

prediction for hexagonal prism is 4π/9√3, the maximum coverage prediction for

pyramid is 5π/12. We can approximate the number of sensor nodes to cover the

sensing field using Prism deployment as

(7)

Number of sensor nodes to cover the sensing field using Cube deployment as

(8)

Number of sensor nodes to cover the sensing field using Hexagonal Prism deployment

as

(9)

Number of sensor nodes to cover the sensing field using Pyramid deployment as

(10)

IV. ENERGY PRESERVING SCHEDULING PROTOCOL

We present a scheduling algorithm which helps us to extend the lifetime of the sensor

nodes with maintaining sufficient coverage. Sensors with overlapping coverage areas

of more than fifty percent are turned off to save energy and are woken up at the

appropriate time to extend the network lifetime. Each sensor implements the algorithm

independently. The sensor can be in any of the four states: Active, Sleep, Idle and

Dead. Each active sensor will try to enter the sleep mode from where after a specific

time interval it goes back to the active mode again. The sensor node can also enter the


idle mode from the active mode after which it enters the dead mode where it is

terminated if it has low energy. The node wishing to enter the sleep mode, first checks

the neighbors whose overlap of sensing area is greater than fifty percent and broadcasts

a sleeping request (SR) message to all neighbors. If all the neighbors agree the node

can enter the sleep mode. If any of the neighbor rejects, the node keeps the trial active

and attempts again after a predefined time. The neighboring node which receives this

request (SR) recalculates the coverage ignoring the requesting sensor. If the coverage

is sufficient then a positive acknowledgment (PAK) is given else a negative

acknowledgment (NAK) is given. Multiple sensors can move to the sleep mode

simultaneously provides an advantage to our algorithm. It is necessary that only one

sensor within the neighborhood is allowed to send a request at a time. Neighbour nodes

randomly contend with each other to avoid collisions. We assume the sensor nodes

whose sensing range overlaps with each other can communicate directly 13. We also

assume the sensor node knows the location of the neighboring sensor nodes. Figure 5

shows the state transition diagram of the algorithm.

Active SleepIdle

Dead

Fig.5. State transition diagram for scheduling algorithm

In the following, we explain how the sensor node works.

i. Active- Initially the sensor node si stays in the active state to sense the environment

and check the neighbor sensor node overlap. It broadcasts the request to sleep message

(SR) to all the neighbors and waits for the reply for a predetermined time using a timer

T1. Each of the neighbors sj’s recalculate the coverage by ignoring the si. If the

coverage is sufficient without si a positive acknowledgment (PAK) is sent else a

negative acknowledgment (NAK) is sent. The three results that can be received for si

request are:

1. All the neighbors send a positive acknowledgment (PAK) the node si goes to

sleep state.

2. If any negative acknowledgment (NAK) is received si stays in the active mode.

3. If the timer goes off before all the replies from the neighbors are received, si

stays in the active mode.


ii. Sleep – The node si broadcasts a confirm message(CON) to indicate to all neighbors

that it is going to the sleep mode and should not be taken into account. It remains in the

sleep state until it is woken by the neighbor sensor nodes.

iii. Idle: The sensor node in the active mode can go to the idle mode when the energy

goes below a threshold level. It starts a timer T2 and sends an Urgent message to all

the neighbor nodes that it is going to die soon and the sleeping neighbor sensor nodes

should be woken up to take over. The active neighbor sensor nodes which have a list

of the sleeping nodes with overlapping coverage send a message to them to get to the

active mode immediately.

iv. Dead/Terminated: The sensor node runs out of energy and terminates.

V. RESULTS AND ANALYSIS

We consolidate our results in this section. We have used Matlab for our simulations.

The simulations show the Coverage Prediction and the Number of Sensors for different

kinds of sensor node deployment. The entire sensing field is assumed to be a cube with

volume V = 500*500*500 m3. The maximum sensing radius Rmax is assumed to be 20

m. The sensing field is partitioned into small equilateral prism, cube, and hexagonal

prism, sub-regions of side 20 m.

The graph in figure 6 shows the sensing radius and coverage prediction for the various

deployments strategies of cube, prism, hexagonal prism, pyramid and random

deployment. We observe that the pyramid type of deployment gives us the maximum

coverage and the hexagonal prism gives the least coverage prediction amongst all the

various types of deployment strategies. The Prism deployment strategy also reaches

very close to maximum coverage prediction of pyramid deployment. Random

deployment as can be seen gives a very random picture of the coverage prediction.

Fig.6 Coverage Prediction vs Sensing radius


Figure 7 represents the sensing radius with the number of sensor nodes required in

each of the deployment strategies of cube, prism, hexagonal prism, pyramid and

random deployment. We can see that the maximum number of sensor nodes are

required by the pyramid deployment followed by prism deployment, cube deployment

and hexagonal prism deployment. Therefore, the pyramid and prism type of

deployment provide good coverage prediction with almost same number of sensors.

Fig.7. Number of sensors vs sensing radius

Fig.8. Coverage prediction vs sensing radius


Fig.9. Coverage prediction vs sensing radius

The above figure 8 represents the coverage prediction for prism and pyramid

deployment since both have the maximum nearly equal coverage prediction. While

coverage prediction for the pyramid deployment comes to around 0.16, prism comes a

close second with 0.15.

The above figure 9 represents the coverage prediction with scheduling. We observe

that with scheduling the coverage in pyramid deployment comes down to about 0.10

from 0.16 and prism deployment comes down further to about 0.08. The difference

with scheduling pyramid deployment in coverage prediction is about 0.06 but

considering the saving of energy that we are having in terms of the number of nodes it

is huge.

The graph in figure 10 depicts the coverage prediction for cube deployment with and

without scheduling. We observe the coverage prediction drops down significantly

when we use scheduling in cube deployment.


Fig.10.Coverage prediction vs sensing radius

The graph in figure 11 depicts the number of sensor nodes required for prism and

pyramid deployment with scheduling. We can clearly see that pyramid deployment

uses the highest number of nodes whereas using scheduling the number of sensor

nodes drops down signficantly to more than half. The number of sensor nodes

required for prism deployment is a tad bit lower than pyramid deployment which

comes down further with scheduling. We interpret that although the deployment of

pyramid and prism is the best which gives us maximum coverage with least number

of nodes if used with scheduling.

Fig.11.Coverage prediction vs number of sensors


The graph in figure 12 shows the coverage prediction for hexagonal prism

deployment with and without scheduling. We observe the coverage prediction drops

down significantly when we use scheduling in hexagonal prism deployment.

Fig.12.Coverage prediction vs number of sensors

VI. CONCLUSION

We have presented the different types of sensor node deployment schemes for 3 D

wireless sensor networks for finding the coverage prediction. Various kinds of

deployment of sensor nodes help to understand the major role deployment plays in the

coverage of wireless sensor networks. We present the prism, cube, pyramid and

hexagonal prism type of node placement. We also give a comparative review of the

various node deployment schemes discussed with the coverage prediction and the

number of sensors used in each case. We also present a scheduling algorithm for the

above discussed schemes and find the difference in coverage prediction and number

of sensor nodes required for each case. We find that pyramid node deployment

scheme has the highest coverage prediction which is almost equal to prism type of

deployment and hexagonal prism has the least coverage prediction. Also although the

pyramid deployment has the best coverage prediction but it also uses the maximum

number of sensors while the hexagon prism uses the least number of sensors along

with the lowest coverage prediction. When we use the pyramid type of deployment

with scheduling we get a very good coverage prediction using less number of sensors.

Therefore most practical deployment scheme is either the pyramid deployment with

scheduling or the prism deployment with scheduling which uses an average number

of sensors for good coverage prediction.


REFERENCES

[1] Potdar V, Sharif A, Chang E. Wireless sensor networks: A survey. In:

Advanced Information Networking and Applications Workshops, 2009. WAINA’09. International Conference on. ; 2009:636-641.

[2] Yick J, Mukherjee B, Ghosal D. Wireless sensor network survey. Comput networks. 2008;52(12):2292-2330.

[3] Wang Y, Zhang Y, Liu J, Bhandari R. Coverage, connectivity, and

deployment in wireless sensor networks. In: Recent Development in Wireless Sensor and Ad-Hoc Networks. Springer; 2015:25-44.

[4] Wang Y, Wang X, Agrawal DP, Minai AA. Impact of heterogeneity on

coverage and broadcast reachability in wireless sensor networks. In:

Proceedings of 15th International Conference on Computer Communications and Networks. ; 2006:63-67.

[5] Ishizuka M, Aida M. Performance study of node placement in sensor

networks. In: Distributed Computing Systems Workshops, 2004. Proceedings. 24th International Conference on. ; 2004:598-603.

[6] Deif DS, Gadallah Y. Classification of wireless sensor networks deployment

techniques. IEEE Commun Surv Tutorials. 2014;16(2):834-855.

[7] Akewar MC, Thakur N V. A study of wireless mobile sensor network

deployment. Int J Comput Networks Wirel Commun. 2012;2:533-541.

[8] Liu X, He D. Ant colony optimization with greedy migration mechanism for

node deployment in wireless sensor networks. J Netw Comput Appl. 2014;39:310-318.

[9] Osmani A, Dehghan M, Pourakbar H, Emdadi P. Fuzzy-based movement-

assisted sensor deployment method in wireless sensor networks. In:

Computational Intelligence, Communication Systems and Networks, 2009. CICSYN’09. First International Conference on. ; 2009:90-95.

[10] Poduri S, Sukhatme GS. Constrained coverage for mobile sensor networks. In:

Robotics and Automation, 2004. Proceedings. ICRA’04. 2004 IEEE International Conference on. Vol 1. ; 2004:165-171.

[11] Katti A, Lobiyal DK. Sensor node deployment and coverage prediction for

underwater sensor networks. In: Computing for Sustainable Global Development (INDIACom), 2016 3rd International Conference on. ;

2016:3018-3022.

[12] Kumar S, Lobiyal DK. Sensing coverage prediction for wireless sensor

networks in shadowed and multipath environment. Sci World J. 2013;2013.

[13] Zhang H, Hou J. On deriving the upper bound of $α$-lifetime for large sensor

networks. In: Proceedings of the 5th ACM International Symposium on Mobile Ad Hoc Networking and Computing. ; 2004:121-132.

node deployment strategies and coverage prediction in 3d wireless sensor network … ·...

Documents